Sustained interruptions are one of the most common power-quality problems, and in the utility industry, they generally fall under the “reliability” umbrella. The problem is that the reliability umbrella is so large that the term can mean different things. A good example of this is reliability with respect to distributed generation (DG).
Almost all of the literature promoting DG also claims that DG improves reliability. However, utility engineers often will not give DG any credit for reliability improvement, and in fact, they often will give reasons why it will decrease reliability.
In seminars that we teach, we conduct a 15-minute exercise during which utility engineers typically will come up with about 20 reasons why a specific DG example does not represent an improvement in reliability. We joke that a good distribution engineer should be able to raise about one barrier to DG per minute.
How can there be such a difference of opinion? It's a matter of perspective and whose reliability you are talking about. One perspective is looking down from the top on the delivery side, the other is looking up from the end-user side. While the utility is primarily concerned with the reliability of the wire-based power-delivery system, DG proponents generally look at the reliability of end users who own and control the generation. These different viewpoints create the controversy.
DG and the Distribution System
The utility-distribution system is designed to deliver power in one direction reliably and inexpensively. Anything that disrupts that initial design has the potential to be detrimental to the reliability. Except in the downtown area of large cities, distribution systems in the United States tend to be simple radial feeders because these are the most economical to operate and protect. When a short-circuit fault occurs, only one fuse or breaker needs to operate to clear the fault. When there is only one source of power, the devices that operate autonomously using local intelligence assume this task. Even low-voltage networks in large cities assume unidirectional power flow in the supply feeders and actually can be more sensitive to DG than radial systems. In either type of distribution system, the introduction of DG in sufficient amounts requires a more sophisticated protection system.
To improve the reliability of the radial power-delivery scheme, utility distribution feeders have multiple sectionalizing switches that allow for reconfiguration of the radial circuits during emergencies, or for balancing loads between substations during normal conditions. The illustration above shows a distribution-planning area fed by four substations. When there is a fault in one section, power can be restored to unaffected sections by changing the switches to isolate the faulted section and pick up unfaulted sections from alternate feeds.
Other devices such as capacitor banks and voltage regulators on the system can complicate matters. Control settings that work well in the normal state do not always work well when switched to the alternate feeds. Manufacturers offer features on the controls to change control strategies automatically when the alternate feed is detected to simplify the restoration process.
Adding DG to this mix creates some additional complexities. One consideration is that the generator must disconnect automatically to allow faults to clear. Another consideration is the interaction between the DG controls and the controls of other elements in the system, which assumes that the system is radial. It is not surprising that utility engineers are concerned about potential problems.
DG to the Rescue?
Adding DG can present reliability benefits, too. For example, consider the case where there is not quite enough capacity to supply the peak load when switched to the backup feeder. Many utilities will build new capacity when this is projected to occur. If the generator is located near the tie between two feeders, it can support the load on either feeder when in the alternate configuration at heavy load (top figure). Thus, utilities can defer investment in wire-delivery facilities by relying on DG, at least for a few years, if not indefinitely.
Although this can be an effective and economic solution, there are some things to consider. First, DG in this location is likely to be owned by the customer. Does the utility really want to rely on someone else's generator to provide capacity? Customers often are willing to use their backup generators for this purpose when offered sufficient capacity credits. These contracts also have significant penalties for failure to perform. To avoid complications from interconnecting the generator, some agreements are written as curtailable load. This accomplishes load reduction for the emergency condition, although more capacity could be used through interconnection.
Not all customer-owned DG will be in an appropriate place to provide such feeder support. Some groups want to treat all DG equally, which may be appropriate for constraints on the transmission grid. However, when the main concern is improving reliability on distribution feeders, the utility must evaluate each location separately and tailor capacity credits for each candidate.
Often the capacity constraint shows up first in feeders out of one substation in an area. Engineers can address this by installing DG in the substation to cover contingencies involving the supply to that substation (bottom figure).
This DG application often is more palatable to utility engineers because it is utility-controlled and operated. The only catch is to make sure that the bus arrangement is sufficiently flexible to allow the DG to be connected as needed for a variety of equipment failures within the substation. Otherwise, there is a limited gain in reliability.
Relaying and Protection
For DG that serves to cover contingencies, utilities may choose not to make costly changes to the protection system because the generation is connected only for a short time after a first contingency. In the event of a second contingency, the suboptimal actions are accepted. Cogeneration is DG that is interconnected most of the time and changes often are made to accommodate it. One common change is directional overcurrent relaying to prevent sympathetic tripping from generator infeed into faults on other feeders. The DG owner is generally charged a fee for making this change. Most are unwilling to pay for doing this for all possible feeds, so interconnected operation is restricted to the normal configuration. If the utility decides to change the normal feed at a later date, it generally will have to absorb any cost for modifying the relays.
Changes made to accommodate DG ripple down to a multitude of operational issues. Crews have to be more careful about taking equipment out of service for maintenance. For example, where does the directional overcurrent relay get its voltage signal? Will the potential transformer be disconnected if the feeder breaker is transferred to another bus?
Using DG to cover capacity shortages during contingencies to defer capital investment in transformers and feeders often has a limited life. The life span of this solution will depend on the load growth, operating life of the generator and any additional measures that might be put in place to meet capacity needs.
Once the new wire-delivery capacity is built, the deferral value of DG is eliminated. It still may have value as a hedge against high power prices and high value as backup power to the customer who owns it. Of course, that will depend on the impact of the investment in wires facilities on the base reliability of the power-delivery system. Sometimes there will be such a dramatic reduction in the number and duration of interruptions that much of the value of backup generation is eliminated.
If the main source of service interruptions is storms, the new wires capacity may not actually affect the number of interruptions a particular customer sees.
This brings to mind one obvious application of DG that does improve the reliability. Many resort loads are remotely located and served from one substation through miles of trees and harsh terrain. Rights-of-way are often inadequate, and aesthetic environmental concerns often preclude the necessary tree trimming. Outages are frequent and long. Short of burying cable, which is very expensive, little can be done to significantly improve the base outage rate. A second overhead line would likely suffer the same outages as the first.
Installing sufficient DG at the resort site to support most of the load during an outage is sometimes a more economical and reliable solution. Such DG can be exploited for demand reduction or supplying thermal loads at the resort.
The Big Picture
Despite these examples, most DG applications will probably have little impact on the reliability of the distribution system as it is presently measured. The DG owner who suffers sufficient power interruptions to justify purchasing backup power realizes the greatest reliability improvement. Utility distribution engineers largely refuse to rely on DG to provide base capacity. They point out that at various times, the wire-based delivery system will have to serve all the load without the help of DG. For example, DG is required to disconnect from the system when there is a fault.
There is far too much distribution infrastructure in place to practically consider changing it just to better accommodate DG. So this is likely to remain a requirement for the foreseeable future. If the system has become dependent on DG to meet the peak load, reclosing becomes a big potential problem. A more complicated procedure would be required to restore service after a fault and outage durations would be expected to increase.
Power-delivery reliability by central generating stations and wires often takes a beating in the popular press. The raw material largely comes from DG advocate groups — including environmentalists, marketers and groups in conflict with utilities — wanting to sell equipment. However if one takes a look at the numbers, wires don't look so bad. Once the wires are in place, they generally remain there quietly doing their duty with relatively minimal maintenance for 40 years or so.
In the United States, one can expect roughly 90 minutes of power outage a year on the average. This works out to a system availability of 99.983%, or almost “four 9s” in the current vernacular. A customer can add another “9” by supplementing this with a single backup generator or ride-through solution such as an uninterruptible power system. In fact, the high reliability touted in some DG literature is achieved in this fashion, by supplementing the base reliability of the wires.
It typically takes four separate generation units to exceed the reliability one can achieve with average utility service. In contrast to wires, nearly all generation technologies require major overhauls at least every five years at a cost of up to 30% of the initial investment. Some require it every other year if they operate most of the time. Of course, many industrial and commercial customers have other items requiring such maintenance and are equipped to handle this. They can more easily take advantage of DG benefits. It is still a major unanswered question whether the average consumer wants to be bothered with owning, operating and maintaining DG.
Most industry experts agree on the need for a more robust and flexible power grid. As retail power markets continue to develop, DG will make a significant contribution.
At that point, distribution systems' control and communications infrastructures must expand. However, the integration of DG must be made with a better understanding of who will benefit. Supporting the distribution system with DG can mutually benefit utilities and customers but can negatively impact reliability. Where and how DG is interconnected determines its value to the system. To show this value, traditional measures of reliability need to be modified. Regardless of the application, connecting DG to distribution systems is a challenge and must be carefully considered from an operational and economic standpoint.
Roger Dugan is senior consultant for Electrotek Concepts (Knoxville, Tennessee, U.S.). He specializes in power-system simulation and analysis, and lately has focused on distributed generation. He is a fellow of the IEEE and a co-author of Electric Power Systems Quality.